In order to investigate the effect of long-term (1991—2013) K fertilizer deficiency and surplus on potassium-bearing mineral and K nutrition of purple soil-soil primary, clay mineral composition and potassium (K) nutrition were determined on the long-term experiment of fertility and fertilizer efficiency in neutral purple soil by using X-ray diffraction (XRD) analysis technique. Five soil samples were selected from soil samples library for soil mineral analysis, including original soil, which preserved in 1991 before the experiment carrying out; K deficient treatment for 12 years and 22 years, which means no K fertilizer was applied during 1991—2003 and 1991—2013, respectively; and K surplus treatment for 12 years and 22 years, which means excess K fertilizer was applied during 1991—2003 and 1991—2013, separately. The result showed that, soil potassium-bearing primary mineral, such as mica, potassium feldspar, had apparently weathered and slaked for the K deficient treatment and the weathered extent gradually aggravated following fertilization ages, demonstrating fertilization for 22 years<12years<original soil. However, for K surplus treatment, the content of mica and potassium feldspar only had a little decline. As similar with primary mineral, for soil clay potassium mineral, the content of illite and mica was also seriously reduced for K deficient treatment and a slightly declined for K surplus treatment, representing the disintegration of soil potassium mineral under rice-wheat rotation system, and the deficient input of K fertilizer would obviously accelerate this weathering process. The two treatments all represented raising of vermiculite with the increasing of fertilization ages, and it’s much higher on K surplus treatment than K deficient treatment. In addition, we found that the silica-sequioxide ration of soil clay, which indicated the degree of soil weathering, decreased as the fertilization ages increasing, and after 22 years K deficient input soil clay had the lowest silica-sequioxide ration, showing the greatest weathering extent of soil clay silicate minerals. To further research the changing of soil K nutrition, we analyzed the content of soil available K, slowly available K and total K. The result indicated that, for K deficient treatment, soil available K and slowly available K content all gradually declined with the increasing of fertilization ages, about 62.0% and 37.4% down from 1991 to 2013, respectively. While there had a gain trend of soil available K content after long-term K surplus input, versus slightly dropping of slowly available K content. However, we found that long-term fertilization had no obvious effect on soil total K content. Therefore, long-term K deficiency could have negative effect on soil K, exhausting soil K resource and accelerating soil mineral weathering. It seems to be difficult maintaining soil slowly available K and potassium-bearing mineral content, even excess applied K fertilizer.